PDA Simulator for CFG Induction Using Genetic Algorithm

Grammar Induction (or Grammar Inference or Language Learning) is the process of learning of a grammar from training data of the positive and negative strings of the language. Genetic algorithms are amongst the techniques which provide successful result for the grammar induction. The paper describes a Pushdown Automata (PDA) simulator used to parse the training data with the grammar induced by the Genetic Algorithm process. The grammar is induced by using an extended approach of stochastic mutation scheme based on Adaptive Genetic. The algorithm produces successive generations of individuals, computing their "fitness value" at each step and selecting the best of them when the termination condition is reached. The paper deals with the issues in implementation of the algorithm, chromosome representation and evaluation, selection and replacement strategy, and the genetic operators for crossover and mutation. The model has been implemented, and the results obtained for the set of four languages are presented.